As will be noted, this balance dispenses entirely with knife edges, and this statement carries with it the gist of its entire merit.  There is no friction, and the elegance of the work and the nice adjustments of the parts struck the writer at once.

[Illustration:  KENT’S TORSION BALANCE.  Fig 5.]

The prescription scale and the proportional scale (see Fig. 4) are particularly interesting.  The former is sensitive to 1/64 of a grain, and the latter, invented by Mr. Kent, is a most ingenious method for weighing, by which, in a small compass (101/2 in. by 41/4 in. by 33/4 in.), we have a balance capable of weighing 3 lb. avoirdupois by thirty-seconds of an ounce.

For ordinary balances on the torsion system, in which extreme sensitiveness is not needed, the trouble caused by change of level of the scale is insignificant; but it becomes a matter of importance in more sensitive scales, such as fine analytical balances in places where it is impossible to keep the table or support of the scale level, for instance on shipboard.

To counteract this effect of the change of level, Dr. Alfred Springer devised the system which is shown in its most elementary form in Fig. 2.  An additional beam, E, with wire, F, and poise, H, on support, C, were added to the balance, and connected to it by a jointed connecting piece, J. The moment of the structure, E C H, about its center of rotation was made equal to the moment of A C D about the center.  The wires, B and F, are attached at their ends to supports which are both rigidly connected to the same base or foundation.  If this base, the normal position of which is horizontal, is tipped slightly, the weights, C and H, will both tend to fall in the same direction.  But suppose the right hand end of the base is raised, causing both of the weights to tip to the left of the vertical, D, tending to fall over, the left tends to raise the right hand end of the beam, and the connecting piece, J H, also tending to fall to the left, tends to lower the left hand end of E and the piece, J. The moments of the structure, E C H, and A B D being equal, and one tending to raise J and the other to lower it, the effect will be zero, and J will remain in its normal position.

It is not at all necessary, however, to have the weights and dimensions of the structure, E C H, equal to those of A B D. All that is necessary is that the components of the weight of each part of the structure which act vertically on J shall be equal and opposite.  For, if the left end of the beam, E, is made shorter than the right end of the beam, A, a given angle of rotation of the beam, A, will cause a greater-angle of rotation of E, consequently will tip the weight, H, further from the vertical than the weight, D, is tipped, and in that case the weight, D, must be made smaller than H, to produce an equal and opposite effect upon J. In practice it is convenient to make the beam, E, only one-fifth to one-twentieth as long as A, and to correspondingly reduce the weight, H, relatively to D. In this case, on account of the angle of rotation of the beam, E, being greater than the angle of rotation of A, the beam, E, becomes a multiplier of the indications of the primary beam, A.